Mining distance-based outliers in near linear time with randomization and a simple pruning rule